Google Owes $811 Billion Whether the Demand Shows Up or Not

TITLE: Google Owes $811 Billion Whether the Demand Shows Up or Not BODY:

Alphabet beat estimates on almost every line in Q2 2026. Revenue up 24% to $119.8 billion. Google Cloud up 82%. The stock fell 7% anyway.

That tells you what the market was actually reading.

The Obligation, Not the Guidance

Alphabet had $811 billion in contracted future spending commitments at the end of June, up nearly $500 billion from three months earlier. These obligations, disclosed separately from capital expenditure, represent purchases already committed under supply agreements and open purchase orders. Guidance can be revised on the next earnings call. A purchase order cannot.

The commitments cover chips, data centers, electricity, and resources for AI infrastructure. Alphabet raised its 2026 capex forecast to $195–$205 billion, reflecting faster computing deployment. Quarterly capex has moved in one direction: $22.4 billion in Q2 2025, $24.0 billion in Q3, $27.9 billion in Q4, $35.7 billion in Q1 2026, $44.9 billion in Q2 2026.

This is a ratchet, not a budget.

Q2 operating cash flow was $39.1 billion. Capital expenditures were $44.9 billion. Free cash flow for the quarter was negative $5.855 billion. For twenty years, Google was the hyperscaler everyone assumed could fund AI out of its own pocket. This quarter, that changed.

Alphabet raised approximately $49.6 billion in equity in June 2026 for AI infrastructure and issued $20.3 billion in senior notes, roughly doubling long-term debt to approximately $98 billion. The company repurchased zero shares this quarter, versus $13.2 billion a year ago.

Translation: Google is now funding the build with equity dilution and debt, because the operational cash engine is no longer large enough to cover the bill.

The 50% Number Cuts Both Ways

Google Cloud CEO Thomas Kurian told CNBC that existing customers are spending "roughly 50% more" than their committed amounts. Kurial also said Google plans to rent third-party capacity for the next few quarters to meet demand, allowing it to onboard customers and transition them to its own infrastructure once capacity comes online. He acknowledged renting this capacity will hurt margins.

Read that sequence. Customers are consuming 50% above commitments. Google cannot serve all of that on its own infrastructure. So Google rents third-party compute at margin-negative rates while locked into $811 billion in obligations to build the capacity that will eventually replace those rentals.

The consensus frames the 50% overage as proof of demand. The correct framing: it is proof that Google's capacity build has not kept pace with consumption.

Google Cloud's backlog reached $514 billion, up more than $50 billion sequentially. That is contracted but not yet delivered revenue. Revenue does not move chips. Capacity does.

What Breaks the Model

CFO Anat Ashkenazi said free cash flow will "remain under pressure" and 2027 capex will "increase significantly" — then declined to put a figure on it. The market had accepted a large 2026 number. Not getting the 2027 anchor is what sold the stock.

Google's bet is one-directional. It works if AI demand compounds at the current rate. Google Cloud grew 82% year over year. Customer acquisition more than doubled. Gemini processing reached about 22 billion tokens per minute. Those are real numbers.

The risk is not that Google's technology stops working. The risk is that demand growth normalizes — slows to 30%, 20%, 15% — while $811 billion in non-cancelable obligations does not. Competitors with more variable capacity can adjust. Google cannot.

A company raising $80 billion in equity while running negative free cash flow is not signaling confidence. It is signaling that demand assumptions must hold.

The chip suppliers get paid regardless. The shareholders absorb the risk if the assumptions break.

What to Watch

Three signals: (1) Whether free cash flow stays negative through Q4 while Cloud revenue growth decelerates below 60%. That combination triggers unwinding questions a $514 billion backlog cannot answer. (2) Any CFO language about "optimizing" supply agreements or "rightsizing" commitments — that is how renegotiation sounds in a press release. (3) The 2027 capex number. Below $200 billion signals the growth bet is being quietly revised downward. Above $250 billion means the obligation schedule grows before revenue catches up.